It started as an April Fool’s Day prank.
On April 1, 2018, researchers at the Massachusetts Institute of Technology (MIT) Media Lab in the United States launched an artificial intelligence (AI) called Norman.
After a few months, Norman, named after the murderous hotel owner in Robert Bloch and Alfred Hitchcock’s Psycho, went on to become the world’s first “psycho AI.”
But Pinar Yanardag and his colleagues at MIT had not built Norman to cause global panic.
It was supposed to be an experiment designed to show one of AI’s most pressing problems: how biased training data can affect the technology’s output.